vnStat [Release] vnStat-on-Merlin - UI, CLI and email - data use and data limit monitoring - R1 and R2

[ColinTaylor]

What does TOOLS > SYSINFO > HW ACCELERATION say about RUNNER and FLOW CACHE (these are the 'new' HwA setting names)?

There are no UI settings for these; the software that requires they be enabled/disabled changes settings in nvram, mostly automatically.

vnStat-on-Merlin does not modify these settings, preferring to leave it to the user to determine if HA is more important than date-use monitoring.

You can modify these settings from the CLI should you choose to do so.

Does runner or flow cache actually have any effect on the accuracy of the stats? I was under the impression that it didn't.

 

[dev_null]

In TOOLS > SYSINFO > NETWORK > HW ACCELERATION, both RUNNER and FLOW CACHE say they are Enabled. What are the CLI commands to disable them? Cake need both to be disabled?

I was under the impression that any QOS (Cake included) required HwA to be disabled. But you may want to check over on the Cake thread. My connection is slow enough to warrant using QOS, so both RUNNER and FC are disabled on my AX86U.

I have not tried this but see here: https://www.snbforums.com/threads/rt-ac86u-hw-acceleration-setting.45375/post-454515

 

[dev_null]

Does runner or flow cache actually have any effect on the accuracy of the stats? I was under the impression that it didn't.

I may be over-simplifying or showing the depth of my knowledge, but it is my understanding that anything that bypasses the TCP/IP stack would potentially impact how the data usage is accounted for. This is what Merlin said about RUNNER and FA few years ago (https://www.snbforums.com/threads/rt-ac86u-hw-acceleration-setting.45375/post-389282):

Runner is the successor to Flow Accelerator. It's a second level of hardware acceleration that is incompatible with a lot of router features.

The most important one is Flow Cache, which replaces the old CTF.

Asus hasn't made either of them user-configurable yet.

So if FC=CTF, that's certainly bypassing the stack. RUNNER, I'm not so clear on.

Others certainly can and will weigh in.

 

[ColinTaylor]

I may be over-simplifying or showing the depth of my knowledge, but it is my understanding that anything that bypasses the TCP/IP stack would potentially impact how the data usage is accounted for. This is what Merlin said about RUNNER and FA few years ago (https://www.snbforums.com/threads/rt-ac86u-hw-acceleration-setting.45375/post-389282):

So if FC=CTF, that's certainly bypassing the stack. RUNNER, I'm not so clear on.

Others certainly can and will weigh in.

Yes I'm aware of that but FC works differently than CTF which is why Adaptive QoS doesn't suffer the same problems as Traditional QoS.

I don't have an HND router so I can't test it. I suggest someone checks this so that people don't make changes that impact performance that might not be necessary.

 

[TonyK132]

I ran runner disable from the CLI. The prompt came back with no errors. But If I go to the UI, I see that runner is still Enabled.
I also ran fc disable. I got this message:
Broadcom Packet Flow Cache learning via BLOG disabled.
After refreshing the GUI, is said both were still enabled.
So what else needs to happen?

 

[dev_null]

I ran runner disable from the CLI. The prompt came back with no errors. But If I go to the UI, I see that runner is still Enabled.
I also ran fc disable. I got this message:
Broadcom Packet Flow Cache learning via BLOG disabled.
After refreshing the GUI, is said both were still enabled.
So what else needs to happen?

Did you nvram commit && reboot ?

 

[TonyK132]

I did not. I did not know to do that. But I just did it based on your comment, and it did not make a difference. The GUI still shows both being Enabled.

 

[Stephen Harrington]

Just merged a hotfix that should address this issue. Thanks to @Jack Yaz

Thanks @dev_null and @Jack Yaz, I'll see what turns up in the logs tomorrow and report.

 

[ColinTaylor]

Did you nvram commit && reboot ?

That will just undo what he just did as he hasn't changed any nvram variables.

 

[dev_null]

That will just undo what he just did as he hasn't changed any nvram variables.

Yeah, I realized that in the middle of the night. Teaches me not to provide advice on a Friday evening after a long, long week.

I did not. I did not know to do that. But I just did it based on your comment, and it did not make a difference. The GUI still shows both being Enabled.

I'd suggest you take this to a separate thread because it's not specifically related to vnStat-on-Merlin.

 

[TonyK132]

Elsewhere in this thread, someone said that these needed to be disabled for vnStat to work and be correct, which is why I brought it up here. But this concept also applies to other apps like Cake.

 

[bluzfanmr1]

Is it possible to merge into the current database a couple of previously exported databases I saved as I was doing a default reset to the router? I ended up not importing them at the time. I've read about --savemerged but haven't been able to find any examples, though I'm not sure that is the correct way to do it.

 

[Stephen Harrington]

Thanks @dev_null and @Jack Yaz, I'll see what turns up in the logs tomorrow and report.

Looks like the spurious mail messages I was getting in the logs each night are now gone with the hotfix, thanks!

 

[dev_null]

Is it possible to merge into the current database a couple of previously exported databases I saved as I was doing a default reset to the router? I ended up not importing them at the time. I've read about --savemerged but haven't been able to find any examples, though I'm not sure that is the correct way to do it.

Not that I have been able to accomplish. I tried to do this using various commands during my initial (pre-alpha) implementation and never succeeded. I even tried to export to a text file, adding the additional data and re-importing, all without success.

 

[dev_null]

Elsewhere in this thread, someone said that these needed to be disabled for vnStat to work and be correct, which is why I brought it up here. But this concept also applies to other apps like Cake.

Actually we don't know definitively how the accounting with vnStat is affected by stack-bypass. While stack-bypass appears to affect the data logging, it most certainly does impact Cake. If you've got Cake installed, it should have modified these settings automatically. I'm not sure why the command line update didn't work for you, but perhaps over in the Cake section they can provide more insight.

 

[pdc]

Oops! I did not do a good job remembering previous threads. This was answered:

This is a unfortunately not something we can control. One or two of the "user-pages" isn't using the underpinning code and will on occasion over-write the menutree or one of the other userX.asp pages.

The way I have found to be solve it is to reboot. I haven't had a time where a reboot (maybe 2) hasn't solved the issue. I know Jack has been trying to get the add-ons codebase aligned to solve this issue, but it's a WIP.

My original post follows...

I posted in the Merlin thread for 386.2, but perhaps this is related to vnStat given the other issues people have had with the add-on pages? Is there a workaround to get the add-on pages back?

Dirty upgrade from _2 to _4, and I am no longer seeing most of my script add-on pages. The Add-Ons page lists only Unbound (user3.asp), but there used to be pages for dn-vnstat, connmon, and scMerlin there as well, but now they do not appear. Per the log they are still there, but something didn't go quite right.

I have tried rebooting a few times but it hasn't helped. The pages are there if I manually type in the user#.asp page.

May  2 19:24:21 AC86U-MAIN dn-vnstat: Mounting WebUI tab for dn-vnstat
May  2 19:24:21 AC86U-MAIN dn-vnstat: Mounted dn-vnstat WebUI page as user4.asp
May  2 19:24:22 AC86U-MAIN connmon: Mounting WebUI tab for connmon
May  2 19:24:22 AC86U-MAIN connmon: Mounted connmon WebUI page as user5.asp
May  2 19:24:29 AC86U-MAIN scMerlin: Mounting WebUI tab for scMerlin
May  2 19:24:30 AC86U-MAIN scMerlin: Mounted scMerlin WebUI page as user7.asp

 

[bluzfanmr1]

Oops! I did not do a good job remembering previous threads. This was answered:


My original post follows...

I posted in the Merlin thread for 386.2, but perhaps this is related to vnStat given the other issues people have had with the add-on pages? Is there a workaround to get the add-on pages back?

Dirty upgrade from _2 to _4, and I am no longer seeing most of my script add-on pages. The Add-Ons page lists only Unbound (user3.asp), but there used to be pages for dn-vnstat, connmon, and scMerlin there as well, but now they do not appear. Per the log they are still there, but something didn't go quite right.

I have tried rebooting a few times but it hasn't helped. The pages are there if I manually type in the user#.asp page.

May  2 19:24:21 AC86U-MAIN dn-vnstat: Mounting WebUI tab for dn-vnstat
May  2 19:24:21 AC86U-MAIN dn-vnstat: Mounted dn-vnstat WebUI page as user4.asp
May  2 19:24:22 AC86U-MAIN connmon: Mounting WebUI tab for connmon
May  2 19:24:22 AC86U-MAIN connmon: Mounted connmon WebUI page as user5.asp
May  2 19:24:29 AC86U-MAIN scMerlin: Mounting WebUI tab for scMerlin
May  2 19:24:30 AC86U-MAIN scMerlin: Mounted scMerlin WebUI page as user7.asp

Multiple reboots didn't work for me either so I uninstalled the Unbound stats gui and rebooted and the vnstat tab was back. Then I reinstalled the Unbound stats gui and both are now there. I haven't rebooted again as I suspect the vnstat tab will disappear again.

 

[pdc]

Multiple reboots didn't work for me either so I uninstalled the Unbound stats gui and rebooted and the vnstat tab was back. Then I reinstalled the Unbound stats gui and both are now there. I haven't rebooted again as I suspect the vnstat tab will disappear again.

That worked! Thanks for the tip

 

[coloxim]

With this script can i disable the original in traffic statistics?

If I use both there more ram/CPU consumption?

Thanks

 

[dev_null]

With this script can i disable the original in traffic statistics?

If I use both there more ram/CPU consumption?

Thanks

As described in the intro, the original intent was to replace the built-in stats because on my RT-AC66U_B1 the stats were unreliable (ghost spikes). A fix pushed by Asus has supposedly fixed this for the HND routers, but some still report the spikes on non-HND models. If you have a HND model, vnStat-on-Merlin probably only adds the 'dashboard' view where you can see all usage without having to select different views.

I do not have a method to disable the built-in stats, but if you do, please share.

My pre-release performance testing with vnStat-on-Merlin demonstrates no material impact on CPU or router performance when run every 5 minutes. This application has been tested on single-core models (N66U) through quad-core (AX86U).

The testing, for example, included the single-core N66U running repeated vnStat-on-Merlin refreshes during video calls and during VOIP calls, with no artifacting noted.